 We're back live at VMworld 2012. This is in San Francisco. This is theCUBE. Siliconangle.com's flagship telecast. We go out to the event, extract a signal from the noise. I'm John Furrier, the founder of Siliconangle.com, and I'm joined by co-host. I'm Dave Vellante of Wikibon.org. We're here with Jay Srivallal, who is the CEO of Arista Networks, a multi-time CUBE guest. Welcome back. Good to always be back. Thank you, Dave and John. It's a pleasure. It's always great to have you on theCUBE because the greatness of Arista and the greatness of theCUBE is something that we love, that combination last year. You were one of our most popular segments, dynamic, very cutting edge, and we were actually talking about how networking's changing, and I'm really excited to have you here today because the networking business is upside down with excitement. People are exploding with interest around networking and software-defined networking. You guys have had a great success in this kind of modern era, as we're calling it, in the data center, leading the charge. And so the first question I got to ask you is, I mean, you've got to be pumped about this software-defined networking trend. Oh boy, it just plays so straight into Arista. First of all, thank you for calling us great. Arista actually in Greek means to be great, so we're working on greatness at all levels, technology people and the people I'm surrounded with right now. So as you rightly say, software-defined networking was literally made for Arista. If you remember last year, I told you networking had gotten sexy, and I think SDN just proves it because what we can now say, and look straight in the eye and say, is we have the only software foundation for networking that's modern, open, and most importantly, programmable. It's this extensibility and programmability that really, really gets to the heart of software-defined networking. And at Arista, we've been working harder on it with a very clean sheet of paper for eight years. And finally, we can see the customers deploying it with much success. Well, I got to say, we talked last year and we had a great CUBE interview. I see out there Pauline, Nis, Hi Pauline from Intel. She actually said on the CUBE, that the network is the bottleneck and we talked about that last year. So Jay Shree, take me through the current state of networking. Obviously, we're bullish on software-defined networking, as you know, and you guys are doing it. How's that bottleneck changing? What are the key trends that people should focus in on when they look at this segment? For the folks that are in IT, not inside the ropes, know the inside baseball, but what are the key things they need to be looking at around this shift to this modern, new networking landscape? Well, as Pauline can attest, I think what's happened is the storage folks have put in terabytes of storage. The compute guys are really pushing multi-cores and multiple compute capacity. And all of a sudden, we always sat back and thought the network was not the bottleneck, but the IO, in fact, is. And the applications and the capability of these applications are pushing the envelope as well. So what you see now is the real emergence of a new type of cloud stack, where no more can you just sort of have good enough for all these technologies, but you really have to build the best of breed for storage, compute, and the network has to, therefore, be a very critical foundation, and literally the IO that connects all of these. And of course, we're at VMworld, so I could not mention the sprawl of virtual machines and how that's affecting networking foundations as well. And so this new combination of best of breed applications transacting over the network and bringing in the right resources of workload, mobility, storage, and compute is, I think, fundamentally changing the way networks are deployed. It just can't be static anymore. Pat Kelsing was on this morning, the new CEO of VMware, first time on theCUBE as the CEO, and we were talking about the virtual machine sprawl. And particularly when you talk about software, you cannot not talk about virtual machines, but specifically how that's impacting VMs up and down the stack. So for example, there's a lot of advances in low level virtual machines that's impacting programmatic decision making, management of programs, big data analytics. So the question I want to ask for you, ask to you is, you've got a unique position. You're a CEO of an emerging company that's at the cutting edge of this, but also you know the tech. So what is your angle on this data-centric infrastructure? Because what you're seeing is data is now at the center of the value proposition, Jay Shree, and we had an earlier comment yesterday around the guys who were talking about operating systems. And in the old days, data wasn't really a part of the operating system design, but now data is, and the quote was, if SSDs were around during OS design, a completely different model would appear. So take us through the view of that. What's going on with this data-centric infrastructure? Is it real? Is it a builds-on conversion network? What's your take on that? If I could add some color to that, John. As well, Jay Shree, for the last five, seven years we've been focused on doing more with less, just to tell you know, TCO, cutting costs. You're talking about new applications. That to me says we're going to drive new business value, really drive productivity for the organization. I think you both are hitting on the massive drivers of why we need to really rethink this problem is the way big data is changing, not just the analytics, but also the applications are no more a closed mainframe or a closed clients of architecture. It's really traffic patterns that are going east-west and pushing the envelope of everything. So let me step back and say, what is it in software that defined Networking that's so exciting and why is big data such a fundamental part of it? First what it says is you can build a more advanced network topology. You can have a leaf spine architecture. You don't have to have layers of high latency. You can bring all of this into one flat topology. This is why the invention by both VMware and Rista of VxLands is so important because we completely remove the concept of subnetting or having layer two boundaries, layer three. All of a sudden you extended the LAN architecture from an artificial limit of 4,000 VLANs to literally unlimited. It can be 16 million segments and addresses. And one spec, one IETS spec that now everybody's embracing is right there. We're showing the industry's first VxLand demo with F5, EMC, Icelon, of course VMware and Rista. But we're bringing all this together wire speed all the way from, like you pointed out, delivering the application to the network reacting to it to all of the big data scale out resources from Icelon responding to all of this. It's pretty amazing when you can do all of this with one technology innovation VxLand. So translate that for me into talk about the business driver of, I got a business, I got people with multiple phones, I got zillions of access points in the network. So. I think you said it best, it's workload mobility of all those zillions of applications and access points but take it another whole step further which is not just across physical boundaries but really physical, to virtual, to cloud. The enterprise is changing. The enterprise is no more a monolithic stack. It's really about bringing the openness and the programmability to every aspect of our resources and applications. And whether it's big data, whether it's web 2.0, whether it's high density compute, or whether it's high frequency trading or massive oil and gas applications or social networking, all of these have some very common attributes that are now requiring networks to respond in a programmatic fashion. So you guys have shipping product in the market with Arista that's cutting edge. We've been seeing on our dashboard, Cisco is putting out a lot of noise in the social sphere around Cisco. I don't even think they actually have a product shipping. And they had an announcement at VMware yesterday. Can you, what was that about? Did you follow that? But it's difficult to parse through all of what Cisco does, how about I tell you what VMware and Arista do and then we can comment on the broader Cisco. Obviously Cisco is a well-established enterprise company, but private and public clouds and especially the movement to software-defined networking, Arista began it with its software and it's the only one in the industry I can challenge anybody who says they have a software-defined network whether it's really an enterprise network that's been sort of funged into a software network or if it truly is open and programmable in every aspect. We've been long partners with VMware for the last three years. We developed VM Tracer, you might know we won the award for VXLAN. This year we demonstrate the industry's first multi-vendor VXLAN demo. What we're really doing, just as VMware brought server virtualization to the industry, together with VMware and the controllers, especially with their NYSERA acquisition, we're really bringing network virtualization to the industry so that we can go beyond an isolated boundary to really the whole world across data centers, across clouds, across virtual industries. You know John, Oracle Bot. And Sam, what was the Cisco thing again? Cisco thing was, what Cisco can best say? Well, so what, you know Oracle Bot is, just say no comment, just say no comment. John, Oracle Bot Zygo, that's called, that's software defined networks, that's SDN. I don't know about that. I don't want to get into Cisco because there's nothing to talk about there because obviously the announcement yesterday was kind of like a, there was really nothing to say. I think it's a confirmation of a partnership they already had, that's good. That's like saying my husband and I are still married. That's the news release, right there. So if the software proliferates, right, the software that you talk about VXLAN, you're changing the role of your company. You're no longer just a switch vendor. You're evolving your company. What are you becoming? I think the best way to describe this is we're really the new standard in software defined cloud networking. We have brought the application to the network and the network to the application and these high performance transactions require a new class, a new dimension and a new more modern software because it requires a new cloud stack and these are no more closed proprietary connectivity points but really a way to enable the next world of workload applications. Jay Shree, let's go, let's get into that because I want to, this is important because there's a lot of fud out there around software defined networking, even the word software gets kicked in there, everyone starts screaming, oh no, I'm this, I'm that. Take us through some use cases, specifically. Yeah, let's talk about what you can really do with software defined networking. The first thing, as I said, is build the right cloud topology. So our use case in many cases is they're moving from the enterprise, they're putting in a big data storage, typically it's a Hadoop cluster and they want to build a large scale out storage and they'll use a Rista leaf and spines and build out the topology, that's one use case. The second is they need to introduce what I'd call SDN control points. So the control points may be for enhancing network virtualization or it may be for quickly rapidly configuring their servers so they can cut down their server configuration time from six hours to 20 minutes. We have a number of large software providers that you're all very familiar with right here in the Bay Area or all over the world that are using zero touch provisioning to do that kind of one touch software defined network. VM Tracer which we work with VMware on is another example of a SDN touch point where you can not only say I'm building a virtual network but I can now transcend virtual to physical as a use case and this is where network virtualization really shines and the final is what you've heard of a lot which is controllers and management. Does the network manage the management station or does the management station manage the network? Well, if you build a very distributed network and the controller has the right open APIs Arista has taken the approach that we can be controller friendly with anyone. We've demonstrated that with vSphere. We can do that with HP or OpenView or some of the service providers build their own management and the key here is to unshackle the APIs, make it programmatic and work with them so you can have a single point of control and manage. So you're talking about top down management model right? Yeah. That's what you're enabling. You're talking about a closed stack now becoming an open stack with best of breed components every step of the way I connect to the best big data storage, the best compute, the best virtualization. So you buy the whole abstract pool automate message? Oh, you bet. And the reason I buy it so much is because it starts with having best of breed server virtualization mapped into your storage and it isn't a single vendor. It's really a multi vendor provided you have the right open architecture and you have well-defined APIs. Okay, so I got to ask you, let's step back for a second. I want to ask you some more of a common question around your take on the Sierra, okay? So here's a startup. I know the guys have met him a few times and they're doing some cool work and open stack, et cetera. You got $50 million in funding, give or take a few million here or there and then boom, a billion dollar acquisition from VMware. I like it, I'm proud of them. So it's good, it's great outcome. But the effect people go, people who aren't in the industry go, oh my God, that's massive. Great exit, everyone does well. What does it mean? So what does it mean? So for the person out there who's trying to read the teeth, why did that happen? Can you share with your opinion on that? One quick analogy first of all, I think NYSERA is like the Instagram of networking. One of the ways to look at that is sometimes and every once in a while there's just awesome technology, awesome team that comes together and they're so valuable, they created a value proposition around not just server virtualization, where VMware is the leader, but really bringing that to the network world, right? And the network virtualization platform, as NYSERA calls it, was really pioneered by some very brilliant folks in NYSERA. And they didn't have, they haven't had the time to take it to market and develop the use cases because they got snapped up earlier. But that doesn't mean there isn't a crown jewel there. And certainly, ERISTA views that crown jewel as NYSERA has built an open virtual switch controller that now complements VMware. And the two together, working with ERISTA EOS and physical switches can transform the way software defined networking. And in particular, network virtualization can be done. So we'll have Martin on. This is technology as its best. So tomorrow at two, we're going to have Martin, the co-founder on, from here, we're going to go into great detail about that great exit and great team and great company. And so it's safe to say then you're excited for ERISTA with this acquisition. It validates us. It validates us, it gives you more room to grow. You think the growth on there will be significant from this? Absolutely. We have been working with NYSERA even before the acquisition and we've been working with VMware a lot. We look forward to seeing more demonstrations together from us. Great. So NYSERA is creating a new land grab around software defined network in which we need and we need more developers. We need more solutions. We had Ping Li on earlier and Frank Artali from the venture community talking about the excitement around the VC community. So question for you is, what kind of new startups might emerge that will come into the space? Because with that kind of exit, obviously it's going to draw some new entrance in. So what's your take on that? Because eight years ago, nobody was funding networking and everybody was saying, Cisco and Cisco's buying every company and that's it. If you actually looked 10, 15 years ago, Cisco had a half a billion dollar market cap and the sum of all the companies couldn't add up to a hundred billion, right? Things have changed today. If you look at the Best of Breed companies, whether it's Riverbed or F5 or Rubber or newcomers like us like Arista, what it says is if you focus on an application segment, if you build great technology and you take it to our customers, our customers are very hungry for alternatives. Our customers are in the center of figuring out their big data strategy, their cloud strategy, their virtualization strategy and the only way to solve that for the VCs is to really innovate there. So what new areas do you think are going to be, is there any white spaces that you see that need to be filled by some startups that you see as opportunities? The gray space is more than white because a lot of people are already filling it, right? But I think the fundamental shift is you're going from just thinking about the stack as an OS and an independent storage layer to really data transactions. The new application is not your standard email or mainframe or client server, but a whole new of data applications that require you to manipulate the data and therefore the software very differently. Whether it's for wired, wireless, mobility, anywhere. So the question I want to ask you here is more about a CIO type. So imagine you're talking to a CIO and you turn your business hat on, you have to kind of talk a little slower. You say, and they ask you, what's this data infrastructure? Because we kind of bought convergent infrastructure. We're modernizing our data center. We're going to the modern era, the Hadoop era, the Arista era. We're moving to the new way. What does it mean? Why is it, what's different about this new data infrastructure that's different from convergent? What would you say to that person? I think in one single word it's agile responsiveness. The CIO has to deliver applications and services in real time, which is more and more difficult to do with an existing legacy infrastructure. So what he or she has to do is peel off and build that private cloud or public cloud and think of it as the Amazon of infrastructure in some ways and really think about the layers and the mission critical applications and not necessarily succumb to their legacy habits and behaviors. So you see a lot of, based on your early comments, a lot of innovation coming to networking. You want to stay independent. I'm inferring that from some of your moves lately. So how do you stay independent? How do you make yourself? I think the best way this can be answered is to say we continue to build best of technology and our customers ask for our independence. As you know, I've shared with you often, we've still been going at the rate of one new customer a day. So over the last three and a half, four years, we have now 1,500 customers and our customers talk to each others and by strengthening our independence if you will, because they're all looking for an alternative and an alternate way of thinking and breaking old habits. You guys do have a great reputation. I go visit a lot of startups and growing startups and web scale companies and I see Arista boxes by the palette in there and they're not buying one or two, they're buying a lot of the high-end boxes. So congratulations on that. Thank you, thanks for your support. I appreciate it. I guess my final question is, what's next for you this year? Obviously the market's frothy. What's your plans? I mean, obviously between can you growth? What's your growth strategy? Well, our business plan is to continue to achieve profitable growth, not just growth and continue to innovate. We're never afraid to build the next new thing even if it replaces the existing thing. So our focus will continue to be big data and application-driven software-defined network continue to power more and more of the clouds. Cloud networking was our mission but it's really taken off now with the cloud providers and service providers because they're seeing that it's mission critical. It's seeing that it's reducing their apex. So we want to be one of those companies that people remember as do as they say and say as they do and really continue to execute on our vision, not just have a vision state. Maybe I'll ask one more question because I have no one more minute left but because you're so great content guest. The VC's out here that was just on, what would you say to them if you ran into them in a meeting, they're like, Jay Shree, what should we be looking for in the next generation software-defined data center? Because now it's not just about software-defined networking, a whole new operating environment is being reconstructed with data infrastructure and big data. How would you guide them? Assuming they're your friends now. They've always were my friends but they're especially friendly now, yes. I would guide them the following. Since your valuation just went up by a billion. I would guide them the following way, which is really look at the problem you're trying to solve whether it's a software-defined storage or compute or virtualization or application problem and don't try to be all things, focus on the micro-segment and do it extremely well because I think that's what's always made VC's innovative, funding innovative engineers and companies and solving a specific problem rather than the broad problem. We are big companies to solve the broader problems. What we've been lacking is the innovation for the specific software-defined issues that have been on the controller or the infrastructure side. Jay Shree, I want to say you're a great cube guest, you're a tech athlete as we say. Thank you. A lot of energy, a lot of knowledge and congratulations on a risk to you guys who continue to do well. I know you've worked hard. When networking wasn't sexy, you started a company, you built a team. The roller coaster of a startup, you've had great success, congratulations and thanks for coming on theCUBE. It's always appreciated, you guys are my favorites. Thank you very much. This is theCUBE at siliconangle.com, we're here at VMworld, we'll be right back with our next guest.